預(yù)測學(xué)生表現(xiàn)、參與度的能力對于研究課題很重要,因?yàn)樗鼈兛梢詭椭處煼乐箤W(xué)生在期末考試前放棄課程,并確定需要額外幫助的學(xué)生。本研究的目的是預(yù)測學(xué)生在在線學(xué)習(xí)課程中會遇到的困難與參與度。我們使用機(jī)器學(xué)習(xí)（ML）算法分析了由稱為數(shù)字電子教育與設(shè)計(jì)套件（Deeds）的技術(shù)增強(qiáng)學(xué)習(xí)（TEL）系統(tǒng)和虛擬學(xué)習(xí)環(huán)境（VLE）記錄的數(shù)據(jù)。Deeds系統(tǒng)允許學(xué)生在記錄輸入數(shù)據(jù)的同時(shí)解決不同難度的電子電路設(shè)計(jì)練習(xí)。VLE從開放大學(xué)（OU）向?qū)W生提供不同的講座、作業(yè)和材料。然后根據(jù)訓(xùn)練數(shù)據(jù)對ML算法進(jìn)行訓(xùn)練,并在測試數(shù)據(jù)上進(jìn)行測試。我們進(jìn)行了k次交叉驗(yàn)證,并計(jì)算了接收機(jī)的工作特性和均方根誤差、召回率、kappa和精度度量來評估模型的性能。結(jié)果表明,與其他算法相比,人工神經(jīng)網(wǎng)絡(luò)（ANN）和支持向量機(jī)（SVM）對在線學(xué)習(xí)過程中學(xué)生學(xué)習(xí)困難的預(yù)測精度較高。此外,研究結(jié)果顯示,決策樹（DT）、J48、JRIP和梯度提升樹（GBT）分類器在預(yù)測VLE課程學(xué)生參與度上表現(xiàn)得更好。神經(jīng)網(wǎng)絡(luò)、支持向量機(jī)、DT、GBT和JRIP可以很容易地集成到在線學(xué)習(xí)系統(tǒng)中;因此,我們希望教師在課程期間根據(jù)相應(yīng)的分析報(bào)告改進(jìn)學(xué)生的表現(xiàn)。 

【文章來源】：上海大學(xué)上海市 211工程院校

【文章頁數(shù)】：114 頁

【學(xué)位級別】：博士

【文章目錄】：
摘要
ABSTRACT
Chapter 1 Introduction
    1.1.Introduction
    1.2.E-learning challenges
    1.3.Importance of the current study
    1.4.The innovation of the current study
    1.5.Current study research questions
    1.6.Contribution
    1.7.Chapter overview
Chapter 2 Background
    2.1 Deeds
    2.2 MOOC and LMS
    2.3 Digital design course
    2.4 Student difficulty in the next session
    2.5 Virtual learning environment(VLE)
    2.6 Student engagement
    2.7 Educational data mining(EDM)
    2.8 Data mining
        2.8.1 Descriptive model
        2.8.2 Predictive model
    2.9 ML techniques used in the current study
        2.9.1.Decision tree(DT)
        2.9.2.J48
        2.9.3.Classification and regression tree(CART)
        2.9.4.JRIP decision rules
        2.9.5.Gradient Boosting trees(GBT)
        2.9.6.Na?ve bayes classifier(NBC)
        2.9.7.Artificial Neural network(ANN)
        2.9.8.Support vector machine(SVM)
        2.9.9.Logistic regression(DT)
Chapter 3 Problem formulation
    3.1 Predict student difficulty in next session
    3.2 Predict student engagement in VLE
Chapter 4 Data description and pre-processing
    4.1 Predict student difficulty in next session
        4.1.1 Data description
        4.1.2 Pre-processing
    4.2 Predict student engagement in VLE
        4.2.1.Data description
        4.2.2.Preprocessing
        4.2.3.Predictors that affect student engagement in web-based system
Chapter 5 Related works
    5.1.Predict student difficulty in next session
        5.1.1.Traditional learning
        5.1.2.Web-based learning
    5.2.Predict student engagement in VLE
Chapter 6 Proposed Methodology
    6.1.Predict student difficulty in next session
        6.1.1.Combination of the predictor variables
        6.1.2.Model training
        6.1.3.Model evaluation
    6.2.Predict student engagement in VLE
        6.2.1.Building and testing the predictive model
    6.3.Performance Metrics
Chapter 7 Experiments and Results
    7.1.Predict student difficulty in next session
        7.1.1.Propose Model adaptability in education
    7.2.Predict student engagement in VLE
        7.2.1.Data visualization and statistical analysis of the data
        7.2.2.Results and discussion
        7.2.3.Development of an engagement prediction system
        7.2.4.OU analysis Dashboard for the current study
        7.2.5.Predictive model application in a web-based system
Chapter 8 Conclusion
References
Published worked
Acknowledgement



本文編號：3367138